The article further explores the history of attempts to replace software engineers, from the creation of the COBOL programming language in the 1950s to modern no-code tools. The author argues that while these tools have partially succeeded, there is still a persistent need for human software developers. The author concludes that while developer software will likely look different in five years due to improved tools, the idea of "AI developers" doing most of the work is more of a necessary marketing message for today's AI tooling startups.
Key takeaways:
- AI developer tools startups like Magic.dev and Cognition Labs are making bold claims about creating AI that can replace human software engineers, with Cognition Labs' product "Devin" reportedly solving 13.86% of GitHub issues unassisted.
- Despite the hype, the author believes these AI tools are more like advanced coding assistants rather than replacements for human engineers, and that they have a long way to go before they can fully replace human developers.
- The author suggests that these startups may be making such outlandish claims out of necessity, as they need to differentiate themselves in a market already saturated with AI coding assistants like GitHub Copilot.
- The author also notes that the idea of replacing human software engineers with automated tools has been around since the 1960s, and while these tools have made programming more efficient, they have not eliminated the need for human developers.